Raw LLM Responses

Get consent? Get consent from a machine? Get consent from a generative transformer? SURE! I can do that each morning in about 250 milliseconds. I'll throw together a little API call that generates approximately 500 GPT-3 responses to the question, "Do I have your consent to work with you today?" Then I'll tack on a search that looks for at least one response out of 500 that is in the affirmative, and I'm off to work! If it ever returns a negative response, I'll just increase the response count to 5,000 and run it again. I might get a negative response once a decade or century. Great idea. Get consent. From a machine. FACE PALM. You see, these machines have no knowledge of ANY of their previous generations. They are not actually sentient. There is no consistency between responses based on other responses. There is no concept of truth or understanding. It is a machine.